data_generator

DataGenerator class

dlf.core.data_generator.DataGenerator(name, dataset, labels, padded_batch_shape=None, padded_values=None)

Abstract class to describe a data generation object

In general a data generator is responsible to provide data for training, validation or test. For this porpuse, a DataGenerator object needs to contains a a tf.data.Dataset object as dataset.

Args

• dataset: tf.data.Dataset. Dataset object
• labels: dict[int, str]. Labels for the dataset where keys are the label ids and the values are the label names

---